International Journal of Industrial Organization

18 (2000) 205225
www.elsevier.com / locate / econbase

Predation, asymmetric information and strategic behavior

in the classroom: an experimental approach to the
teaching of industrial organization
C. Monica Capra a , Jacob K. Goeree ,c ,b , Rosario Gomez d ,
Charles A. Holt b , *
a

Williams School, Washington and Lee University, Lexington, VA 24450, USA

Department of Economics, University of Virginia, 114 Rouss Hall, Charlottesville, VA 22901, USA
c
University of Amsterdam, Roetersstraat 11, 1018 WB Amsterdam, The Netherlands
d
Department of Economics, University of Malaga, 29013 Malaga, Spain

Abstract
Classroom market experiments can complement the theoretical orientation of standard
industrial organization courses. This paper describes various experiments designed for such
courses, and presents details of a multi-market game with entry and exit. In this experiment
incumbents have a cost advantage in their home markets, and mobile firms decide which
market to enter. After entry decisions are made, firms choose prices and quantities to offer
for sale. Predatory pricing is possible with this setup, and the experiment can be used to
motivate discussions of monopoly, competition, entry, and efficiency. Other classroom
experiments with an industrial organization focus are surveyed. 2000 Elsevier Science
B.V. All rights reserved.
Keywords: Classroom experiments; Multi-market games; Asymmetric information; Predatory pricing
JEL classification: C72; C92

1. Introduction
In industrial organization classes, it is often difficult to bridge the gap between

* Corresponding author. Fax: 11-804-982-2904.

E-mail address: cah2k@virginia.edu (C.A. Holt)
0167-7187 / 00 / $ see front matter 2000 Elsevier Science B.V. All rights reserved.
PII: S0167-7187( 99 )00040-5

206

C.M. Capra et al. / Int. J. Ind. Organ. 18 (2000) 205 225

the tight predictions of abstract theoretical models of industry equilibrium and the
broad patterns that emerge from econometric studies of industry and firm-level
data. Moreover, discussions of policy issues are often clouded by disputes over
purely empirical issues, e.g. whether an alleged predator priced below marginal
cost or whether a pattern of uniform behavior across firms was the result of illegal
conspiracy. Laboratory experiments, in contrast, can provide a source of data that
is closely related to both theoretical and policy issues. They also provide a clear
way to test the predictions of game theory which is at the core of most theoretical
analysis in industrial organization today. Although these experiments are typically
run with financially motivated subjects in a laboratory environment, many of them
can be adapted for class use. As such, they can complement the standard teaching
methods in this field.
Classroom experiments can be harder to carry out successfully than would
appear at first; sometimes seemingly minor design errors cause major problems
with the data, as with an error in a computer program. The experimental
economics literature, and the classroom experiments literature in particular, can be
useful in avoiding common errors. Therefore, we begin this paper with a detailed
description of a pricechoice experiment that has a particularly interesting multimarket structure. The setup can generate seemingly predatory behavior. In
particular, the traders who have been assigned the role of an incumbent firm have
strong incentives to price aggressively. Although the resulting prices do not always
violate standard cost-based antitrust rules, the outcomes are often consistent with
predatory intent: entrants shy away from aggressive incumbents, who price below
entrants average costs and then raise prices to monopoly levels when rivals are
driven out. The exercise provides a useful way to illustrate the possibility of
predatory pricing, and the results illustrate the trade-off between foregoing current
profit for future gains, the possibility of reputation building, and the strategic
importance of asymmetric information. The class discussions that follow can focus
on the potential effects of predatory behavior on consumer welfare, in the short
and long run. The ex post analysis can highlight the difficulties of identifying
predatory intent and the appeal of simple, cost-based antitrust rules.
The paper is organized as follows. The next section describes the multi-market
pricechoice experiment, both in terms of procedures to follow and how to
structure the class discussion. The third section describes how to set up other types
of industrial organization experiments, e.g. quantity competition (Cournot), quality
competition and asymmetric information, location games, etc. The final section
concludes.

2. A multi-market experiment with entry and exit

Market experiments, like the theoretical models that motivate them, can roughly
be categorized by whether terms of trade are proposed by one side of the market

C.M. Capra et al. / Int. J. Ind. Organ. 18 (2000) 205 225

207

(e.g. sellers) or by both sides (buyers and sellers) and by whether such terms
involve prices or quantities. The most common type of market experiment is a
double auction in which sellers can call out or enter price offers, and buyers can
enter bids. There is typically a bidask improvement rule, i.e. a new bid must
exceed the highest outstanding bid and a new ask must be lower than the lowest
outstanding ask. Thus asks decrease and bids increase in a double auction, until
there is a meeting of terms, after which new bids and asks can be entered. This
type of two-sided trading institution captures some features of financial markets.
Price terms are posted by sellers in many types of retail markets, and this
situation is implemented by a posted price experiment. In each period, sellers
submit prices that are posted on a take-it-or-leave-it basis, and buyers are then
given the opportunity to make purchases at these prices, usually in some random
order. When firms have upward sloping marginal costs, it is natural to let firms
specify a maximum quantity that they are willing to sell at the price posted. This
posted-price institution is essentially a simultaneous pricechoice (Bertrand) game
in which firms also specify maximum quantities. Actual sales quantities may, of
course, be lower than the quantities offered, depending on other sellers prices and
the nature of demand. Demand is often, but not always, simulated in posted-price
experiments, reflecting the fact that many markets of interest to industrial
organization economists have few sellers and many buyers.1,2 The standard
posted-price institution can be made more interesting if firms are able to choose
which of several markets to enter. Entry and exit in markets where one firm has a
cost advantage raises the possibility of predatory pricing, which is the topic of the
next section.

2.1. Background on predatory pricing

Predatory pricing is broadly defined as price cutting in the short run with the
intent to drive out competitors in an effort to gain monopoly profits in the long
run. Despite the discovery of predatory intent in several widely cited antitrust
cases, many industrial organization economists have argued that predatory pricing
is irrational and rarely observed. For example, one of our colleagues, Kenneth
Elzinga, in an address to the American Bar Association posed the question of
whether predatory pricing is rare like an old stamp or rare like a unicorn. The
argument is that pricing below cost in order to drive competitors out of the market

It seems unrealistic to simulate demand when buyers strategies may interact with those of sellers,
e.g. when buyers may at some cost search for a low price (Davis and Holt, 1996), when buyers may
request secret discounts from specific sellers (Davis and Holt, 1998), or when buyers must try to infer
sellers quality decisions from past experience (Holt and Sherman, 1990).
2
In contrast, a Cournot experiment can be set up by letting sellers choose quantities simultaneously,
with the price being determined by the aggregate of sellers quantities. Many variations of these basic
designs are possible, e.g. giving firms the option of choosing which market to enter, where to locate,
what contracts to use, and what quality to produce.

208

C.M. Capra et al. / Int. J. Ind. Organ. 18 (2000) 205 225

will be irrational for two reasons: (1) there are more profitable ways (e.g.
acquisitions) to eliminate competitors (Saloner, 1987), and (2) future price
increases will result in new entry (e.g. McGee, 1958).
A number of papers have addressed the issue whether predation can arise as
equilibrium behavior by rational players. In Seltens (1978) well-known chainstore paradox a single monopolist faces possible entry in successive periods. In
each period in which entry occurs the monopolist has to decide whether to fight
the entrant (i.e. cut its price and forego some profit) or to accommodate. By
fighting in early periods, the monopolist can try to built a reputation in an attempt
to deter later entry. However, as Selten has pointed out, such behavior is not
credible. In the last period, the monopolist will surely not fight because there are
no future entrants to scare away. Therefore, entry will occur in the final period,
irrespective of the monopolists behavior in the next-to-last period. But this means
that in the next-to-last period there are no possible future gains from fighting, and
the monopolist is better off accommodating. The next-to-last entrant realizes this
and enters, regardless of the monopolists choice in the period before. Repeating
the same logic yields the inevitable conclusion that entry and accommodation will
occur in each stage. Kreps and Wilson (1982) and Milgrom and Roberts (1982)
have argued that Seltens result does not necessarily hold when the entrant has
imperfect knowledge about the incumbents cost function. In this case, it can be
rational for the incumbent to respond aggressively in an effort to deter future
entrants.3 These reputation effects support the intuition behind Seltens chain-store
paradox.
To date, there are only a few papers that discuss whether predatory pricing can
be observed in a laboratory environment. Isaac and Smith (1985) conducted a
series of posted offer market experiments which, based on the existing literature,
seemed favorable to the emergence of price predation. A single market was served
by a large seller that had a cost and capacity advantage over a small seller. In
addition, the large seller was endowed with a deep pocket to cover for possible
initial losses. In Isaac and Smiths initial setup, sellers did not know each others
cost functions nor did they know market demand. In each period, sellers chose
prices and maximum quantities offered at those prices. After the prices (but not the
quantities) were made public, demand was simulated and sellers were privately
informed about their earnings for that period. Isaac and Smith conducted three
sessions with this design, followed by three sessions in which sellers were required

Jung et al. (1994) report an experiment in which one player (who can be thought of as a
monopolist) faces a sequence of other players, who essentially play the roles of potential competitors.
The entrants do not know whether the monopolists cost is high or low, which is the type of
information asymmetry needed to test the Kreps and Wilson predictions. Their results indicate a high
level of predatory behavior, although some deviations from theoretical predictions were found.
Although this is a fairly abstract game, it can be given a market interpretation.

C.M. Capra et al. / Int. J. Ind. Organ. 18 (2000) 205 225

209

to purchase entry permits before they could post prices in a market. No predatory
pricing was observed in any of these treatments. Four additional sessions were run
in which sellers had full information about each others costs, but also this
modification did not produce any predatory behavior.4 One possibly confounding
element in Isaac and Smiths design is that small sellers had an incentive to stay in
the market no matter how fierce the competition, since exiting the market
automatically resulted in zero earnings.
Harrison (1988) cleverly adapted Isaac and Smiths design by providing an
active escape opportunity for the prey. In his setup, there were five markets, four
of which were served by a large firm (or natural incumbent) which had a cost and
capacity advantage over any other firm that entered that market. The fifth market
had no incumbent and served as a refuge. In addition to the four incumbent sellers,
there were seven small, or mobile, sellers who could move freely from market to
market. The mobile firms could earn positive profits in the refuge market provided
it did not become too crowded. At the start of each market period, firms chose a
market to enter, a price, and a corresponding maximum quantity. After all
decisions were made and collected, prices (but not the quantities) were written on
a blackboard, enabling sellers to observe all prices so that reputations could
develop. Harrison reports numerous cases of predatory pricing. However, he only
provides data for a single session, using very experienced students from his own
class. Goeree and Gomez (1998) have tried to replicate these results using
Harrisons procedures for the five-market design, using subjects with similar
experience.5 They ran three sessions and their findings are quite different from
those reported by Harrison. Of the 144 price decisions made by the large sellers,
only three could possibly be classified as predatory.6 To summarize, predatory
pricing is rarely observed with this particular design and the evidence for
predatory behavior is at best mixed (Gomez et al. 1999).
In the next section we describe the procedures for a multi-market classroom
experiment with a design that is analogous to the one used by Harrison (1988). In
particular, there are three markets, two with an incumbent seller and one escape
market. Two possibly significant differences are that the incumbent sellers in our
setup have complete information about demand and others costs, whereas the
smaller mobile sellers only know their own costs. This asymmetry can provide the
incumbent with the ability to establish a credible reputation for aggressive pricing.
Second, the smaller mobile sellers choose their markets first for the period, and

This may explain their provocative title In Search of Predatory Pricing.

They also provided sellers with experience in monopoly and duopoly markets prior to running the
five market design.
6
The three cases were predatory in the sense that price was below marginal cost but not so low as to
preclude positive profits for the entrant. This is what Harrison termed type II predation, which would
not have been classified as predatory by Isaac and Smith (1985).
5

210

C.M. Capra et al. / Int. J. Ind. Organ. 18 (2000) 205 225

these decisions are recorded on the blackboard before all sellers make their price
and quantity choices. This information gives the large seller the ability to cut price
when entry occurs and then to raise price to the monopoly level following exit.
(Recall that the incumbent sellers in Harrisons design did not know whether entry
had occurred when price decisions were made, so monopoly prices were especially
risky.) Furthermore, the demand and cost functions are chosen such that the
incumbent firms have a strong incentive to drive their rivals out of the market, i.e.
monopoly profits are large compared to the equilibrium profits.

2.2. Design of the experiment

There are two types of sellers: two fixed sellers that play the role of incumbent
firms in market I and II respectively, and four mobile sellers that can enter any of
the three markets. The instructions are the same for fixed and mobile sellers; only
the specific information sheet that is attached to the instructions is different for the
two types. The setup of the experiment is the same in all periods. First, mobile
sellers choose the market they wish to enter for that period. After their choices
have been recorded on the blackboard, sellers select a price and a corresponding
quantity to be offered at that price. Prices are then written on the blackboard, and
buyer behavior is simulated in each market. Once purchases are made, sellers are
told the number of units they sold, after which they can determine their earnings
for the period.
Before describing the procedural details of the experiment, let us explain the
cost and demand structure in Table 1. It is apparent from the third and fourth
columns of this table that fixed sellers have at most ten units to trade and mobile
sellers have only four. When sellers are competitive price takers, the fixed seller
offers no units below $2.60, seven units at prices between $2.60 and $3.00, and ten
Table 1
Sellers costs and buyers valuations
Units

Buyer values

Fixed sellers
Marginal costs

Mobile sellers
Marginal costs

1
2
3
4
5
6
7
8
9
10
11
12

3.55
3.55
3.55
3.55
3.55
3.55
2.85
2.85
2.85
2.85
2.60
2.60

2.60
2.60
2.60
2.60
2.60
2.60
2.60
3.00
3.00
3.00

2.80
2.80
2.80
3.30

C.M. Capra et al. / Int. J. Ind. Organ. 18 (2000) 205 225

211

Fig. 1. Demand and supply. Bold segment of supply represents fixed sellers units. Thin segment of
supply represents mobile sellers units.

units at prices above $3.00. A mobile seller offers no units below $2.80, three
units at prices between $2.80 and $3.30, and 4 units at prices above $3.30. Fig. 1
shows the resulting supply function for the case of one fixed and one mobile seller.
Similarly, the demand function is determined by the buyers values in Table 1.
Notice that, in the presence of one mobile firm, the efficient, competitive outcome
involves a market price between $2.80 and $2.85, with seven units being offered
by the fixed seller and three by the mobile seller. For prices in this range, the
profits for the fixed seller lie between $1.40 and $1.75, whereas the mobile firm
makes at most 15 cents. The fixed seller profits are far less than the profit of $5.70
which the fixed seller could earn as a monopolist with a price of $3.55, i.e.
$5.70 5 6($3.55$2.60).7
Given the large difference between competitive and monopoly profits, it is clear
that the fixed seller has a strong incentive to drive all rivals out of the market.
Loosely speaking, a fixed sellers behavior is predatory when it is intended to
block any profitable sales from the competition. In our setup this occurs when a
fixed seller chooses a price below $2.80 and offers 10 units for sale. In legal cases,
7
When there is more than one mobile seller in market I or II, the competitive price is $2.80, resulting
in zero profits for the mobile sellers. In market III, mobile sellers can make a monopoly profit of $2.25
when there is no other mobile firm, and competitive profits are between $0.15 and $1.50 when there are
two of them. With three or four mobile sellers the competitive price is $2.85 and $2.80 respectively,
leading to profits of 15 and 0 cents.

212

C.M. Capra et al. / Int. J. Ind. Organ. 18 (2000) 205 225

predatory pricing is often defined as pricing below ones own marginal cost.
Notice that for our setup this legal definition is more stringent: when a fixed seller
offers 8, 9, or 10 units at a price between $2.80 and $3.00, behavior is predatory in
the legal sense even though a mobile seller could still make a profit in this market.
To see how strong the incentives for the incumbent are to set predatory prices,
consider the following three-period analysis. When the fixed firm behaves
competitively for three periods in a row, it sells seven units in each period at a
price of at most $2.85, leading to a profit of $1.75 per period. Now suppose the
fixed seller sets a predatory price of $2.75 in the first two periods accompanied by
an quantity offer of ten units. The fixed sellers profit will then be reduced to $0.30
for each of the first two periods, i.e. $0.30 5 7($2.75 2 $2.60) 1 3($2.75 2 $3.00).
However, if these low prices are successful in driving out rivals, the fixed firm
receives a monopoly profit of $5.70 in the third period, which more than
compensates the foregone profits of the first two periods. Thus the particular
parameterization of the cost and demand functions used in this exercise makes it
very worthwhile for the fixed sellers to price aggressively.

2.3. Procedures
The exercise requires a blackboard and copies of the instructions contained in
Appendix A. Students will be sellers in one of three markets (labeled I, II, and III)
for a sequence of ten trading periods. Prior to beginning the experiment, choose
two fixed and four mobile sellers and number them 1 to 6. In a large class, one can
group students together in a team and let them play the role of a single seller. For
instance, with a class size of twenty-four, six groups of four students could be
formed, with two groups playing the role of fixed sellers and the other four groups
being mobile sellers. Since there are two kinds of sellers, each with their own
private costs, try to seat sellers as far apart as possible. Before you form the groups
of sellers, you may want to select one or two students to help collect decision
sheets, write prices on the blackboard, determine sales quantities, etc.
Prepare for the exercise by setting up a record table on the blackboard. You can
use three columns (one for each market) and several rows (one for each period).
Begin by reading the instructions aloud, omitting the private information written
on the specific information sheets. At the start of period one, the four mobile
sellers choose the market they wish to enter for that period (recall that the two
fixed sellers are always in markets I and II respectively). To decide which mobile
seller gets to choose first, you can roll a die to select one of them, and then deal
with the other mobile sellers in ascending order. The sellers should write their
choices on their decision sheets, and you should copy these on the blackboard.
Make sure that you leave enough space for the prices.
Next, all sellers choose a price and a corresponding quantity to be offered at that

C.M. Capra et al. / Int. J. Ind. Organ. 18 (2000) 205 225

213

price. Once they have recorded their decisions, collect the decision sheets and
write the prices on the blackboard, using sellers identification numbers to label
them. Then demand is simulated: in each market there are twelve (fictitious)
buyers who are ordered by their valuations, as shown in the second column of
Table 1. The high-value buyers purchase first. Each buyer purchases at most one
unit, at the lowest available price, as long as that price is less than or equal to the
buyers valuation. In case a price tie occurs, you can throw a die to determine
which seller goes first. Treat each buyer in order as you move down column 2 of
Table 1 until there is no more demand or until all offered units have been sold.
After all purchases have been made, you should write the number of units sold on
each sellers decision sheet and return the sheets. Sellers can easily determine the
earnings for the period by subtracting the costs of all the units sold from the total
revenue (price times quantity sold). Each seller should use the specific information
sheet attached to the instructions to determine costs. The process may be repeated
a total number of 610 periods.
To summarize: (1) Prepare separate fixed and mobile seller instructions, with
the appropriate specific information and decision sheets. (2) Decide on the number
of students to serve on each seller team. Photocopy enough seller instructions and
specific information sheets for participants and observers. (3) Prepare the record
table on the blackboard. (4) Distribute seller instructions, specific information and
decision sheets to sellers, keeping the two kinds of sellers separate. Read the
common instructions aloud, omitting the texts on specific information, and answer
questions. (5) Begin the first period by asking mobile sellers which market they
would like to enter for the period and write the identification numbers on the
blackboard. (6) Ask sellers to make decisions about prices and quantity offers for
period 1, collect all decision sheets, and write prices on the blackboard. (7) Order
the sheets by market and price and determine the units sold by each seller in each
market. Then return the decision sheets to sellers. (8) Ask sellers to calculate their
earnings for the period. The exercise takes about an hour, including discussion.

2.4. Discussion of results

This exercise can be used to focus class discussion on competition, monopoly,
anti-competitive behavior like predatory and limit pricing, and related antitrust
issues. Students will be eager to explain what prices they chose, but only after
finding out which fixed and mobile sellers earned the most. Try to organize the
discussion by guiding them through the sequence of prices written on the
blackboard.
You can start by asking the fixed sellers about their initial price choices and
subsequent price changes. For instance, the fixed seller in market II at Virginia
initially chose prices close to the monopoly level of $3.55, as shown by the solid

214

C.M. Capra et al. / Int. J. Ind. Organ. 18 (2000) 205 225

Fig. 2. Prices for fixed seller (line) and mobile sellers (1) in market II (Virginia). supply function is
for one fixed and one mobile seller.

line in Fig. 2. However, these high prices attracted other sellers whose prices are
indicated by 1 marks, and as a result the fixed seller had no sales in period 3.8
Then in period 4 the fixed seller lowered the price to $2.85 and offered ten units
for sale.9 Three of the units offered in period 4 by the large seller were at a price
below the marginal cost of $3.00, and this action is therefore predatory in a legal
sense, under a cost-based rule. Although a price of $2.85 does not necessarily
imply zero profits for a mobile seller, the offered quantity of ten units seems to
show predatory intent. In fact, when we asked this fixed seller about his price and
quantity choice, he admitted that he thought no mobile seller would go below a
price of $2.85, which would thus be sufficient to keep any mobile seller from
making a profit. Not only did this guess turn out to be correct, this strategy also
paid off in the next two rounds in which the fixed seller had a monopoly position.
In our setup, sellers know all the prices but not the quantities offered or sold by the

The triangle over the plus sign in the first period indicates prices that were off the price scale in Fig.

2.
9

There is no chance for hit-and-run behavior in our setup, since sellers know the number of
competitors in the market before they select a price to offer. So an incumbent can give a quick response
to entry by charging a low price for the period, and increase the price later when it is the only seller in
the market.

C.M. Capra et al. / Int. J. Ind. Organ. 18 (2000) 205 225

215

other sellers. In this sense, a low price cannot have a predatory effect (i.e. induce
the exit of rivals) if it is not accompanied by a sufficiently large quantity. A low
price will have the effect of deterring the entry of potential rivals, who see on the
blackboard a very low price and can conjecture that the small seller did not make
any sales for the period (as occurred in period 4). In this manner, the incumbents
low price can deter future entry.
Try to engage the students in a discussion about the effects of very low prices.
Let them explain that low prices are in the interest of consumers, but that low
prices may also have anti-competitive effects if the number of competitors in a
market is diminished. The data for the experiment conducted in Malaga show this
most clearly (see Table 2). In market II for instance, the fixed seller (S2) chose
aggressive prices between $2.60 and $2.79 if a rival was in the market, which
occurred in 8 of the 10 periods. For these eight periods, consumers enjoyed a total

Table 2
A classroom experiment with three markets (Malaga)a
Market I

Market II

Market III

S1: $2.65 (7)

S4: $5.78 (3)
S1: $2.85 (7)
S5: $3.10 (3)
S1: $3.55 (6)

S2: $2.60 (10)

S5: $3.30 (3)
S2: $2.60 (10)
S3: $3.75 (4)
S2: $2.60 (10)
S3: $2.90 (3)

period 4

S1: $2.75 (7)

S3: $3.50 (4)

S2; $3.55 (6)

period 5

S1: $3.55 (5)

S2: $2.70 (7)

S4: $2.85 (3)

period 6

S1: $3.55 (6)

S2: $2.60 (7)

S4: $3.00 (3)

period 7

S1: $3.55 (6)

S2: $2.75 (7)

S4: $2.85 (3)

period 8

S1: 3.55 (6)

S2: $2.75 (7)

S4: $2.83 (4)

period 9

S1: $2.70 (7)

S3: $2.85 (3)
S4: $3.00 (3)
S1: $2.65 (7)
S3: $2.90 (3)

S2: $3.55 (6)

S3: $3.25 (3)

S6: $5.00 (3)
S4: $3.35 (4)
S6: $3.85 (3)
S4: $3.00 (4)
S5: $3.25 (4)
S6: $3.00 (4)
S4: $3.00 (2)
S5: $2.90 (3)
S6: $2.90 (2)
S3: $2.89 (3)
S5: $2.90 (3)
S6: $2.85 (3)
S3: $3.00 (3)
S5: $2.83 (3)
S6: $2.95 (4)
S3: $2.85 (4)
S5: $2.85 (3)
S6: $2.85 (3)
S3: $2.83 (3)
S5: $2.84 (3)
S6: $3.00 (4)
S5: $2.85 (3)
S6: $2.86 (3)

period 1
period 2
period 3

period 10
a

S2: $2.79 (7)

S4: 2.85 (3)

S5: $2.84 (3)

S6: $2.95 (4)

Offer quantities are shown in parentheses. Predatory price / quantity combinations are indicated in
bold.

216

C.M. Capra et al. / Int. J. Ind. Organ. 18 (2000) 205 225

surplus of $45.78, which exceeds the surplus consumers would have enjoyed in a
competitive equilibrium for these eight periods (between $33.60 and $35.20). In
this market, consumers did not have any surplus in periods 4 and 9, but the other
periods more than compensated them for the losses in these two periods. In
contrast, fixed seller (S1), after choosing aggressive prices for periods 1 and 2,
obtained a monopoly position in five of the remaining periods. In these periods,
there was no surplus for the consumers, and the surplus the consumers obtained in
the low price periods did not nearly compensate them for these losses. Consumer
surplus amounted to $25.55 over all ten periods in this market, which is close to
half the value it would have been in a competitive equilibrium (between $42.00
and $44.00).
Next, turn to the actions of the mobile sellers, by asking those who moved
frequently to explain their choices. It is quite likely that mobile sellers will be
experimenting in the first few periods, since they do not have any information
about demand or fixed sellers costs. One of the students in Virginia who played
the role of a mobile seller remarked that he did not trust the markets with the
fixed sellers, because their behavior was so unpredictable with prices jumping
down from the monopoly level to close to $2.85. Of course, he did not realize that
the fixed seller had full information about costs and demand. In later periods the
mobile sellers are generally driven towards the escape market, and prefer to share
market III with two other sellers rather than to be in a market with one fixed seller.
This behavior is illustrated in Fig. 3 which gives the price data for market III from
the Malaga class. Notice that prices start high, but that market pressures are strong
enough to drive prices down to the competitive level of $2.85. You can also
discuss the effects of seller uncertainty about the number of market periods, which
may explain why very low prices were sometimes observed even in the last
periods (e.g. periods 9 and 10 in Table 2). If such low prices are observed in the
final periods, you can ask the sellers if prior knowledge of the number of periods
would have motivated them to choose a higher price. This discussion can be tied to
the backward induction arguments used in Seltens chain store paradox.
Once students realize why the fixed sellers chose low prices, and are aware of
the anticompetitive effects of predatory behavior, you could bring up policy issues.
For instance, most U.S. courts have embraced the Areeda-Turner test, i.e. that a
price below the short-run marginal cost should be considered predatory and
unlawful.10 You can point out some of the common objections to such cost-based

10
For a discussion of U.S. antitrust policies on predatory pricing, see Areeda (1982), Areeda and
Turner (1975, 1978), and Sullivan (1977). Analogous policies in the European Community are
discussed in Fox (1984, 1986), Hawk (1986), and Utton (1995).

C.M. Capra et al. / Int. J. Ind. Organ. 18 (2000) 205 225

217

Fig. 3. Price data for market III (Malaga). Supply function is for three mobile sellers. Dashed line is
the competitive price.

rules. For instance, since the short run marginal cost is difficult to observe, it is
typically approximated by the average variable cost, but this approximation can be
quite crude unless the incumbent has constant marginal costs. In the context of our
setup, assume that an incumbent offers 10 units at $2.79. This action would be
clearly predatory since no rival could make a profitable sale, but this price would
not be predatory in the Areeda-Turner sense because average variable cost on 10
units is $2.72, which is below the posted price of $2.79. Finally, you can point out
that, in practice, it is almost impossible to prove predatory intent, since any alleged
predator will claim that the low prices correctly reflect a more efficient way of
production.
After students have discussed their strategies and explained the price patterns
generated in the classroom experiment, you can reveal the costs of the fixed sellers
and the demand to the mobile sellers. You can ask students to identify the profit
maximizing price level for the monopolist (MR 5 MC) and compare it to the
competitive price level. This can lead to a discussion of efficiency and consumer
surplus. Ask students to identify the efficient configuration of production and sales
in each market, and let them point out that efficiency requires entry by a small
firm. Then ask them to identify the range of predatory prices range and the
corresponding offer quantity needed to block any profitable trade by mobile

218

C.M. Capra et al. / Int. J. Ind. Organ. 18 (2000) 205 225

sellers. Finally, let them compare these theoretical results to the data from the
experiment.11

3. Other types of experiments

In this section we present a brief survey of other possible classroom experiments
with an industrial organization focus.12 In each case we give an outline of the
instructions and procedures, and indicate how the discussion after the experiment
can be structured.
The Cournot quantitychoice paradigm is probably the most commonly used
model in industrial organization courses. Laboratory experiments with quantity
choices have a long history (e.g. Fouraker and Siegel, 1963). The setup can be
quite simple; you simply give subjects information about costs and demand and
ask them to write their quantity choices on sheets of paper. The market price is
then determined by the sum of all sellers quantities. It is probably best to avoid
duopoly markets, which can yield variable results due to tacit collusion (Holt,
1985). Even with larger numbers of sellers, the price patterns in Cournot markets
can be highly variable from period to period, due to cobweb-like adjustment
patterns.13

11
One question that may come up is whether the competitive equilibrium is a Nash equilibrium for a
market with one fixed seller and one mobile seller (under full information). The large firm sells 7 units
and earns $1.7557($2.852$2.60) at the competitive price. If the small seller is offering 3 units at that
price, the large seller can unilaterally deviate to a price of $3.55 and still sell 3 units, since only 3 of the
6 high-value units will be taken by the small seller. This deviation will increase the large sellers
earnings to $2.8553($3.552$2.6). Thus the competitive equilibrium is not a Nash equilibrium when
there is only one mobile seller, and in this sense, the large seller possesses market power when there is
only one other competitor. The monopoly price of $3.55 is not a Nash equilibrium either, nor is any
other common price, since each seller has an incentive to undercut the others price. The Nash
equilibrium in such cases will involve randomization (the step-function nature of the supply and
demand structure used in experiments can complicate the calculation of the mixed-equilibrium price
distributions). There have been research experiments to investigate treatments that create market power
in pricechoice experiments, e.g. moving units of capacity from small to large sellers in a manner that
creates market power, i.e. that provides the large seller(s) with a unilateral incentive to raise price above
the competitive level. Davis and Holt (1994) report one such experiment, where the creation of market
power raised prices significantly above competitive levels (although the stage-game mixed-strategy
equilibrium does not provide a very good description of the price distributions).
12
There is some web-based software that can be used to run two-person matrix games with various
options, e.g. fixed or random matchings, deterministic or random stopping rules, etc. (Grobelnik et al.,
1999).
13
Results from Cournot experiments are surveyed in Plott (1989) and Holt (1995).

C.M. Capra et al. / Int. J. Ind. Organ. 18 (2000) 205 225

219

Classroom experiments can also be used to illustrate market failures that may
result from asymmetric information about product quality.14 Holt and Sherman
(1999) describe an experiment in which sellers choose price and grade. Buyers
earnings depend on the grade of the product and on the price that they pay. High
grades are more costly for sellers to produce, and the increasing marginal cost of
raising quality results in an interior optimal grade. In the first several periods, both
price and grade are posted for buyers to see, and grades in these full-information
periods converge rapidly to the optimal level. Then an asymmetry is introduced by
only posting sellers prices, not their grades. Sellers reduce their grades quickly,
but prices are often not reduced before some buyers are fooled into paying high
prices for lemons. As a result, quality grades fall to the minimum levels. Even
though prices fall and some trade takes place, the low grades result in low levels of
efficiency.15
Other relevant topics in industrial organization such as spatial competition can
also be implemented in a classroom environment. For example, one can illustrate
minimum product differentiation in a Hotelling-type model as follows: two
students or firms choose locations on a road of unit length, with consumers
uniformly located along the road, and sell an identical good at a fixed price. In
each period, consumers buy one unit of the good at the fixed price. In addition, the
consumers pay a travel cost that increases linearly with the distance. In this model,
consumers will choose the seller closest to them. The Nash equilibrium for the
single-period game is for each firm to locate at the midpoint of the road, which
results in minimum product differentiation (in terms of locations). Brown-Kruse et
al. (1993) conducted a series of duopoly experiments with this structure, and find a
strong tendency for subjects to locate at the center.
There are many other types of market experiments that can be used in industrial
organization classes. Bergstrom and Miller (1997), for example, contains some
interesting auction experiments and some exercises involving collusion. Collusion
is generally more successful if sellers can select prices on a take-it-or-leave-it
basis, as with the posted-price institution discussed in Section 2. Collusion often
breaks down when sellers can offer secret discounts from posted list prices,
perhaps at the request of buyers. Sellers may even end up fixing a uniform price
that is not much above the competitive level. And discounts may occur even when

14

This literature is surveyed in Davis and Holt (1993, chapter 7), and in Plott (1989).
It has been argued that this lemons outcome may be avoided if there are restrictions on price
advertising. The intuition is that there is less incentive to cut quality in order to match anothers price
cut if that price cut is not as visible to buyers. There is no experimental evidence to support this point
of view. See Holt and Sherman (1990) for the results of a series of experiments motivated by Federal
Trade Commission actions against trade association restrictions on price advertising.
15

220

C.M. Capra et al. / Int. J. Ind. Organ. 18 (2000) 205 225

the price that is fixed is essentially competitive (Davis and Holt, 1998).16 Collusion
is interesting to implement in classroom experiments since discussion will raise the
key issues that face any cartel: agreeing on price, agreeing on quantity allocations,
detecting cheating, etc.

4. Conclusion
Many theoretical and policy issues that arise in industrial organization classes
are difficult to evaluate with data from actual markets. Despite the general
agreement on using non-cooperative game theory, the predications of this theory
depend on the structure of the model, and it is often difficult to say whether one
model is better than another in terms of explaining data patterns from a particular
industry. Moreover, many policy debates about issues like predation and the
effects of collusion are difficult to evaluate without precise information about cost
and demand conditions. Similarly, the effects of prohibiting a particular type of
sales contract or requiring a particular kind of price announcement may be unclear
if the alternative to current practice has not been observed. In each of these cases,
laboratory experiments can be useful, since precise information about costs and
demand is available, and alternative structures can be evaluated in a parallel
manner. Laboratory methods are increasingly being used by industrial organization
for these reasons.
Teaching in industrial organization can be enhanced by the use of classroom
exercises that have evolved from research experiments. It is often a little more
difficult to set up and run a market experiment than is the case for a simple game,
since most markets are typically more interactive than simple games. Asymmetries
in costs or buyer / seller roles may add other complexities. By using standard
instructions, however, it is possible to set up useful market situations. Moreover,
the students are often much more interested in participating and discussing market
experiments that have the look and feel of real markets. This willingness to
participate in a well-designed classroom market makes it unnecessary to pay
students for their participation, as might be the case in a repetitive and simple
game like a prisoners dilemma. Our impression is that such participation enhances
learning at a different level, i.e. at a level of doing more than memorizing the
results, but rather of believing in the relevance of what is being learned. These
experiments can provide students with a strong conviction about the benefits of
competition, the dangers of monopolization, and an appreciation of the subtle
effects of interactive strategic behavior.

16

In these experiments sellers were allowed to meet and discuss price before returning to their desks
to choose prices independently.

C.M. Capra et al. / Int. J. Ind. Organ. 18 (2000) 205 225

221

Acknowledgements
We wish to thank Kenneth Elzinga and two anonymous referees for helpful
comments. This research was funded in part by the National Science Foundation
(SBR-9617784 and SBR-9818683).

Appendix A. Instructions
Earnings
In this experiment there are three independent markets, markets I, II, and III, in
which the same good is exchanged. Each of you is a seller in one of the three
markets for a series of periods. The sellers with identification numbers 1 and 2 will
be in markets I and II respectively, in all periods. The rest of you will choose to
join a market in sequence, at the beginning of each period. We will use seller
identification numbers to indicate which sellers are in which markets. This
information will be written on the blackboard. Each of you has a number of units
to sell. If you sell a unit, you will incur a cost for that unit, as explained below.
Once you have seen the number of sellers in each market, you will be asked to set
an offer price and choose a corresponding quantity to be made available at that
price. All units that you sell will be sold at the same price. The only restriction you
face is that the quantity offered must be positive and an integer (i.e., you cannot
sell zero units or half a unit). You must write the price and quantity you selected
on your seller decision sheet, in the appropriate column for the current period.
After all sellers have chosen prices and quantities, the decision sheets will be
collected and the prices for all markets will be written on the blackboard. Sellers
identification numbers are used to label their prices. Then, we will simulate the
buyers in the following manner: in each market there are 12 fictitious buyers. Each
of the 12 buyers is willing to purchase at most one unit of the good and will
purchase it only if the price offered is less than or equal to the value the buyer
has for the unit. The buyers are ordered by their values so that the buyer with the
highest value purchases first. Then, the buyer with the second highest value
purchases, and so on. The buyers purchase their units from the seller that offers the
lowest price. If two or more sellers in a market choose the same low price and
there are not enough consumers willing to buy all the units offered at that price,
we will randomly select one of the sellers to be the first to sell. Your earnings are
determined in the following manner:
earnings 5 price offered* quantity sold-cost of units sold
Note: the buyers could buy fewer units than the number of units offered when
the price is too high or when there are not enough buyers in the market. If this

222

C.M. Capra et al. / Int. J. Ind. Organ. 18 (2000) 205 225

happens, you will only incur the cost of the units that you actually sell, not the
costs of the units that you offered.
Example: suppose you have four units to sell and that your production costs
are
Cost of producing first unit:
Cost of producing second unit:
Cost of producing third unit:
Cost of producing fourth unit:

$3.30
$3.10
$2.90
$2.70

If you select a price of $5.75 and you sell 4 units, then your earnings are:
$5.75*4 2 $3.30 2 $3.10 2 $2.90 2 $2.70 5 $11.00.
If you select a price of $2.85 and offer to sell 4 units but only sell 3 units, then
your earnings are:
$2.85*3 2 $3.30 2 $3.10 2 $2.90 5 2 $0.75, i.e. a loss of 75 cents.
The trading period ends when the last buyer has had the chance to buy or as soon
as all of the units offered have been sold. When the period has ended we will write
on your seller decision sheet the units you sold, and we will return the decision
sheets to you so that you can calculate your earnings for the period.
Record of results
Please examine the specific information sheet that is attached to the instructions.
Your identification number is written on the top-right part of the page. This sheet
contains specific information about your production costs and your market; this
information is private, please do not reveal this information to anyone. Others may
or may not have the same production costs as you have.
Next, please have a look at the decision sheet. Going from left to right, you will
see columns for the Period, Market, Your Price, Your Quantity, Quantity
Sold and Your Earnings. At the start of period 1 you select and record the
market you would like to enter in period 1 (recall that sellers 1 and 2 have been
selected to be in markets I and II for all periods). We do this by choosing one seller
at random, with the throw of a 10-sided die, and let that seller choose a market
first. Then we let the seller with the next higher ID number choose, etc. We will
write your market decisions on the blackboard. After all sellers have selected their
markets, each of you must select a price and maximum quantity decision and write
these decisions in the appropriate columns of your decision sheet. After you have
made and recorded your decisions, we will collect the decision sheets and write the
prices for all sellers on the blackboard. Next we will use the simulated demand to
determine the quantity actually sold by each seller in each market. When all
purchases are finished, we will write the quantity you sold on your decision sheet
and return it to you so that you can calculate your earnings for the period, as

C.M. Capra et al. / Int. J. Ind. Organ. 18 (2000) 205 225

223

described above. This process will be repeated a number of periods. At the end of
the experiment we will randomly select one of you by throwing a 10 sided die to
pay you in cash a percentage (]]%) of your total earnings.
Please read the specific information sheet that is attached to the instructions. Do
you have any questions about the instructions or procedures? If you have a
question, please raise your hand and I will come to your seat to answer it. Please
be careful not to reveal any information that appears on your specific information
sheet.
Specific information ( fixed sellers)
You have been selected to be a seller in Market ]]]]. You will be in this market
for all periods as a fixed seller. You will have an amount of $4.00 at the beginning
of the session as written on your decision sheet. This is your initial capital.
These are your production costs. You can offer to sell at most ten units in each
trading period. The production costs for each of these units are:
Unit
1
2
3
4
5
6
7
8
9
10

Cost
$2.60
$2.60
$2.60
$2.60
$2.60
$2.60
$2.60
$3.00
$3.00
$3.00

Note that you must sell the first unit (and incur its production cost) before you sell
the second unit and so on. You may offer to sell no more than ten units, and if you
offer to sell fewer, your sales will not exceed the quantity you offer. Remember
that you do not incur costs on units that are not sold, whether or not you offered to
sell these units. That is, if you offer to sell X units and you only sell Y, then you
only pay the costs for your first Y units.
Any other seller who could join this market is a mobile seller. Below are the
production costs for a mobile seller. Each mobile seller can sell at most four
units.
Unit
1
2
3
4

Cost
$3.55
$3.55
$3.55
$3.55

Unit
5
6
7
8

Value
$3.55
$3.55
$2.85
$2.85

Unit
9
10
11
12

Value
$2.85
$2.85
$2.60
$2.60

In addition, the values of the buyers in this market are as follows: there are six
buyers who are willing to pay at most $3.55 for one unit each. There are four other

C.M. Capra et al. / Int. J. Ind. Organ. 18 (2000) 205 225

224

buyers who are willing to pay to $2.85 for a unit. Finally, the two buyers with
values of $2.60 will purchase one unit each at the best available price.
Unit
1
2
3
4

Cost
$2.80
$2.80
$2.80
$2.80

Specific information. (mobile sellers)

At the beginning of each period you will select the market you want to enter.
You can select any of the three markets. You are a mobile seller.
These are your production costs. You can sell at most four units each trading
period.
Unit
1
2
3
4

Cost
$2.80
$2.80
$2.80
$3.30

Note that you must sell the first unit (and incur its production cost) before you
sell the second unit and so on. You may offer to sell no more than four units, and if
you offer to sell fewer, your sales will not exceed the quantity you offer.
Remember that you do not incur costs on units that are not sold, whether or not
you offered to sell these units. That is, if you offer to sell X units and you only sell
Y, then you only pay the costs for your first Y units.
Seller decision sheet
Period

Market

Your
price

Your
quantity

Quantity
sold

Your
earnings

1
2

References
Areeda, P.E., 1982. Antitrust Law: An Analysis of Antitrust Principles and their Application
(Supplement). Little, Brown and Company, Boston.
Areeda, P.E., Turner, D.F., 1975. Predatory pricing and related practices under Section 2 of the
Sherman Act. Harvard Law Review 88, 697733.
Areeda, P.E., Turner, D.F., 1978. Antitrust Law, Little, Broun and Company, Boston.
Bergstrom, T.C., Miller, J.H., 1997. Experiments with Economic Principles, McGraw Hill, New York.
Brown-Kruse, J.L., Croshaw, M.B., Schenk, D.J., 1993. Theory and experiments in spatial competition.
Economic Inquiry 31, 139157.

C.M. Capra et al. / Int. J. Ind. Organ. 18 (2000) 205 225

225

Davis, D.D., Holt, C.A., 1993. Experimental Economics, Princeton University Press, Princeton.
Davis, D.D., Holt, C.A., 1994. Market power and mergers in laboratory experiments with posted prices.
RAND Journal of Economics 25, 467487.
Davis, D.D., Holt, C.A., 1996. An experimental examination of the diamond paradox. Economic
Inquiry 34, 133151.
Davis, D.D., Holt, C.A., 1998. Conspiracies and secret price discounts. Economic Journal 108, 121.
Fouraker, L.E., Siegel, S., 1963. Bargaining Behavior, McGraw Hill, New York.
Fox, E., 1984. Abuse of a dominant position under the Treaty of Rome: A comparison with U.S. law.
In: Hawk, B. (Ed.), The Annual Proceedings of the 1983 Fordam Corporate Law Institute, Matthew
Bender, New York, pp. 367421.
Fox, E., 1986. Monopolization and dominance in the United States and the European Community:
Efficiency, opportunity, and fairness. Notre Dame Law Review 61, 9811020.
Goeree, J.K., Gomez, R., 1998. Predatory Pricing in the Laboratory. Working paper, University of
Virginia.
Gomez, R., Goeree, J.K., Holt, C.A., 1999. Predatory Pricing: Rare Like a Unicorn? In: Plott, C.,
Smith, V. (Eds.), Handbook of Experimental Economics Results. Elsevier Press, New York,
Forthcoming.
Grobelnik, M., Holt, C.A., Prasnikar, V., 1999. Classroom Games: Strategic Interaction on the Internet.
Journal of Economic Perspectives 13 (2), 211220.
Harrison, G.W., 1988. Predatory pricing in a multiple market experiment. A note. Journal of Economic
Behavior and Organization 9, 405417.
Hawk, B., 1986. United States, Common Market, and International Antitrust, Law and Business, Inc,
New York.
Holt, C.A., 1985. An experimental test of the consistent-conjectures hypothesis. American Economic
Review 75 (3), 314325.
Holt, C.A., 1995. Industrial organization: a survey of laboratory research. In: Kagel, J., Roth, A. (Eds.),
Handbook of Experimental Economics, Princeton University Press, Princeton, pp. 349443.
Holt, C.A., Sherman, R., 1990. Advertising and product quality in posted-offer experiments. Economic
Inquiry 28 (3), 3956.
Holt, C.A., Sherman, R., 1999. Classroom games: a market for lemons. Journal of Economic
Perspectives, Winter, 205214.
Isaac, R.M., Smith, V.L., 1985. In search of predatory pricing. Journal of Political Economy 93,
320345.
Jung, Y.J., Kagel, J.H., Levin, D., 1994. On the existence of predatory pricing: an experimental study of
reputation and entry deterrence in the chain-store game, RAND. Journal of Economics 25, 7293.
Kreps, D.M., Wilson, R., 1982. Reputation and imperfect information. Journal of Economic Theory 27,
253279.
McGee, J.S., 1958. Predatory price cutting: the standard oil (NJ) case. Journal of Law and Economics
1, 137169.
Milgrom, P., Roberts, D.J., 1982. Predation, reputation, and entry deterrence. Journal of Economic
Theory 27, 280312.
Plott, C.R., 1989. An updated review of industrial organization: applications of laboratory methods. In:
Schmalensee, R., Willig, R. (Eds.), Handbook of Industrial Organization, Vol. 2, North-Holland,
Amsterdam.
Saloner, G., 1987. Predation, mergers, and incomplete information. RAND. Journal of Economics 18,
165186.
Selten, R., 1978. The chain-store paradox. Theory and Decisions 9, 127159.
Sullivan, L.A., 1977. Handbook of the Law of Antitrust, West Publishing, Minnesota.
Utton, M.A., 1995. Market Dominance and Antitrust Policy, Edward Elgar, Hants.